I also need to take into account the standard errors of the estimates. The Hausman test accounts for both differences in the estimated parameters and their standard errors. In this case, it confirms that I should reject the assumptions of the RE model (with a large chi-squared value and low p-value) and use the FE estimates.